The final piece of the puzzle: 2008,Q3
Fasten your seatbelts; it could be a very interesting next few days. I will get to all the comments from yesterday, but first there is breaking news that I found quite exciting (when you hear it you’ll probably be thinking: “Exciting? Get a life.”)
First a bit of background. When I started the blog in February I focused on the sharp collapse in the economy after the failure of Lehman in mid-September. I argued that the stock market crash of early October 2008 reflected a loss of confidence in the Fed’s willingness and/or ability to maintain adequate NGDP growth going forward. But many readers found this “errors of omission” argument unconvincing. Monetary policy looked accommodative, and it seemed implausible that it was the Fed’s fault for not immediately staunching the bleeding from the financial crisis. Surely the causation went from financial crisis to falling AD. Perhaps the Fed didn’t do enough to offset this shock, but (so the argument went) they can hardly have caused the problem.
A few months later I discussed 7 pieces of evidence that monetary policy was effectively tightening even before Lehman failed, and the economy was weakening. But even here there were two big problems:
1. The phrase “effectively tightening” is only convincing to those who buy into my forward-looking view of monetary policy. The Fed didn’t actually seem to do anything contractionary, like raise the fed funds target.
2. Yes, individual sectors such as industrial production began a sharp decline in August, and the housing downturn also got much worse around that time, but as this USA Today story shows, real GDP declined only 0.3% in the 3rd quarter. That’s not great, but it’s not much different that the 2007.4 or 2008.1 numbers. So why did the financial crisis get much worse in the 4th quarter?
In the past few days I have received answers to both of these questions. The Hetzel article I linked to earlier has an excellent discussion of how Fed policy went off course in the all important third quarter of 2008. I plan an in depth discussion of that evidence in a few days. But last things first. It is much easy to answer the second question, and the answer is about as clear and unambiguous as you can imagine.
When I read the initial news reports on the 2009 second quarter GDP this morning, I was intrigued by a comment that the BEA had also sharply revised the data for 2008 downward. I had been puzzled by the fact that so many individual sectors did very poorly in 2008:Q3, and yet the overall number wasn’t that bad. Is it possible that the revisions would be concentrated in that pivotal quarter? Take a look at the real GDP numbers in this revised GDP report. The new third quarter RGDP number was revised from negative 0.3% to negative 2.7%.
I should mention for those that don’t follow national income accounting closely (in other words normal people), that this number is worse than it looks, indicative an economy than is doing poorly. One has to remember that unlike highly cyclical sectors like industrial production, GDP includes a huge services component (health, education, government, etc) that is relatively stable. So if RGDP goes from the normal 3.0% growth to a negative 2.7%, you are in a serious recession. Yes, the next two quarters were much more horrific (around negative 6%) but that is very atypical of postwar US recessions. Also note that 2008:Q3 was the worst quarter of the recession for consumption. (I wonder if that led to inventory depletion in Q4—the data is not reported.)
The bottom line is that we now know that the economy got much worse before the financial crisis worsened (post-Lehman.) And the study by Hetzel strongly suggests that Fed policy errors during the 3rd quarter went beyond mere “errors of omission.” All the pieces are now in place. We have a plausible story of how excessively tight money in the third quarter led the a worsening of the financial crisis in late September, and also how the Fed’s failure to aggressively ease monetary policy in early October (focusing instead on ineffective bank bailout programs) caused asset prices to crash and accelerated the economic slump.
My next post will discuss how Milton Friedman’s policy ideas were influenced by Robert Hetzel. I mention it now in case non-economists reading this post are tempted to dismiss Hetzel as some minor figure who can safely be ignored. I also plan a very controversial piece on inflation and the first of many China posts in the lead up to my trip.
Tags:
31. July 2009 at 08:22
Hindsight is always 20/20. I hope you are not blaming the FED because I remember when they first lowered interest rates by the full percent instead of the half everyone was expecting, there was mass pandemonium about how this would lead to massive inflation. You are damned if you do and damned if you don’t.
31. July 2009 at 08:54
Sounds like a slam dunk–a classic stagflation. Inflation expectations adjusted upward. Such that we had high inflation overall but low inflation compared to expectations and probably not enough room in the FF target to compensate.
31. July 2009 at 09:49
Hmm, that’s a large adjustment (same with Q1 09). I’m sure there will be conspiracy theorists arguing that this number was deliberately manipulated to avoid frightening people. Pretty much the same folks who argue that China stopped reporting the electricity consumption data because it was revealing the deception in the official GDP numbers…
As to Q3 08, it’s important to remember that this was coming off of Q2 08. And Q2 08 was goosed by the stimulus rebate checks (which is why we see it break the negative trendline from Q1 08). So a chunk of that decline might have been mean-reversion from the artificially high Q2 numbers. Also, a substantial amount of the real decline occurred at the end of September (post 16th). So if the question was, how bad was the economy in August 08, the -2.5% number might be overstating the case.
As I recall, the federal government was feeling pretty good in June of 08 that it had pulled through the recession with the aid of the stimulus, and now the Fed had to keep a lid on things and we’d have a soft landing. Remember, by May, the Dow Jones had hit 13k again (from its september peak of 14k), which is pretty amazing for being 2+ quarters into a recession…
A lot of people have been worried about a double dip recession. Arguably, we’re already HAD a double dip recession…
31. July 2009 at 11:05
NGDP revisions take six months. As other posters noted, implicitly, this means that central banks will likely always be about six months late in policy, a number that Mark Thoma quoted in one of his early research papers. We are bandwidth limited and there is not much central banking can do to overcome bandwidth limits.
A competitive monetary system may perform better.
31. July 2009 at 11:11
Paul, Most people felt the Fed did far too little. If they were worried about inflation I don’t think you would have seen the Federal government enact an 800 billion dollar stimulus program. That program was done precisely because most people (including Bernanke himself), thought inflation was likely to be far too low, even after the interest rate cuts.
Remember that easy money and fiscal stimulus have exactly the same objective–faster NGDP growth.
Jon, It was very different from the stagflation of the 1970s. In the 1970s you had near double digit increases in wages and prices, and very high interest rates. Oil prices were soaring. In the third quarter oil prices were falling sharply, and interest rates were very low. Five year inflation expectations had fallen to 1.23% right before the fateful September 16 meeting. That’s falling AD, not falling AS.
Statsguy, Yeah, you’re right that some will cry conspiracy. I really don’t have time for conspiracy theorists. I’ve known people that work for the BLS, it’s just routine work. I’ve also met people who don’t think a plane hit the Pentagon on 9/11. Life’s too short to try to convince people like that.
The other thing that is so childish about those theories is the idea that people would see the banking system falling apart, the stock market collapsing and then walk into a voting booth and vote for McCain because the government said RGDP only fell 0.3% in the 3rd quarter. I’ve never met a non-economist who even pays much attention to those numbers.
The Chinese case is interesting. Local leaders are rewarded partly on the basis of the growth numbers—so there might be some padding. On the other hand when you visit Chinese cities they look much richer that the government stats show, so who knows?
Lehman occurred much too late to have any significant effect on Q3 output, I don’t think there is much debate about that. Factories don’t change their production schedules the instant a bank gets into trouble–and remember that the stock market crash didn’t occur until October, so initially investors did think Lehman would be a huge problem (although there was some concern.)
I don’t see a bounceback down from Q2, because I think Q2 reflected monetary ease in January, not the stimulus (which had only a small effect.
You said:
“So if the question was, how bad was the economy in August 08, the -2.5% number might be overstating the case.”
That’s really my point. We already had a lot of data from other sources (industrial production, real estate, etc) that things started getting a lot worse in August. Originally the GDP numbers conflicted with that. Now they are consistent with what we already suspected.
I also see this as a double dip recession, a small initial dip, a slight upswing in Q2, and then a massive second dip.
31. July 2009 at 13:29
Mattyoung, That is exactly why we need a market-based, forward-looking monetary system. Markets were signaling money was way too tight in September 2008, even though the GDP numbers weren’t in yet. They should follow the markets, not out of date data.
31. July 2009 at 14:22
Scott,
Your post raises an important question about the implementation of NGDP targeting. How do we get to measure whether forecast NGDP is consistent with realisations, given the possibility of revisions? Do we need an annual GDP target that gives us a sufficiently long horizon to work out any possible revisions to GDP growth during the quarters of that particular year?
I think your post also validates some of the points you’ve made about NGDP futures targeting – backward-looking monetary policy (whether price level, inflation, or NGDP) can be problematic and subject to “lagged effects”
Alan
31. July 2009 at 16:19
Scott: You keep obsessing over CPI and measures of expected CPI. ‘Inflation’ in the sense of economic theory is a much more abstract concept. If I’m a heavy industry management my expectations of inflation are shaped by the cost of my inputs and the cost of my outputs. I can point to many many industries where those metrics were off the charts and ‘inflation’ expectations reset to very high-levels. By Q3, disinflation started to roll in. Short-terms rates started looking very tight to those people.
31. July 2009 at 17:03
Jon: “‘Inflation’ in the sense of economic theory is a much more abstract concept. If I’m a heavy industry management my expectations of inflation are shaped by the cost of my inputs and the cost of my outputs. I can point to many many industries where those metrics were off the charts and ‘inflation’ expectations reset to very high-levels”
I agree with what you are saying about the nature of inflation.
However, I’m not sure I agree with your view of inflation at the time you mention. It’s not so simple, many industries buy through long term contracts. I agree that an increase in uncertainty occurred.
31. July 2009 at 17:49
Alan, In my futures targeting proposal I suggested that the contracts not be paid off until the second revision was in, at least 14 months after the contracts were traded.
That’s a good point about the problem with backward-looking policies, the data isn’t that reliable. It’s like a fogged up rear view mirror. Again, markets can be wrong, but looking down the road while steering at least gives us a better chance than looking backward.
Jon, Ironically, my next post says we should stop obsessing over the CPI, as it tells us nothing. So I favor NGDP, but NGDP growth was very low in 2008, so I don’t consider that a “problem” in the sense of the 1970s, when NGDP growth was high. I realize many industrial commodities rose in price, but we had to let that happen as otherwise we would have killed the economy (which we eventually did anyway.) We’re now in a world economy, the huge Asian boom was pushing up commodity prices, it wasn’t because our economy was over heated and it wasn’t because US monetary policy was too loose.
31. July 2009 at 19:09
And yet part of the rise in commodity prices can be traced to the steep decline in the exchange-value of the dollar. In this sense, I think its fair to say that policy had been loose in early 2008.
Asian demand was a factor. Supply disruptions was a factor. A weakening dollar was a problem.
1. August 2009 at 04:55
Jon, I just don’t agree that money was loose in early 2008. I think the low rates reflected the steep drop in the demand for credit after the sub-prime bubble burst. If money had been loose, you would have seen high NGDP growth, but if fact it was very low throughout 2008.
1. August 2009 at 08:09
Yes, but that’s the stagflation that I’m talking about. We’ve been over this before–I believe that real-growth was negative at the time.
The difference between now and 1970 is that we have an open economy amid a much more developed world–which is recipe for getting high inflation.
So lets try to compromise here: US policy was ‘relatively loose’–relative to the ROW which is enough to get very high commodity inflation.
2. August 2009 at 08:38
The economy got much worse pre-Lehman because of the housing bust. The national average of housing prices peaked in 2007, which began to cause problems for household and bank balance sheets. The Dow was also at 14,000 and the S and P over 1200 at the time, so the idea that market player were expecting low NGDP in the future is unconvincing, as the nominal value of equities was high. It’s easy to say, what if, but without housing, you are leaving out the major piece of the puzzle. What made Lehman and Bear Stearns fail? expectation of low NGDP? No, losses related to housing.
The Fed didn’t cause asset prices to crash, the popping of a bubble did.
2. August 2009 at 11:24
Jon, Ok, I see your point, but we are still talking about something like 2% NGDP growth, as compared to the normal 5%. So you can call that stagflation, and there is a bit of truth in that. But since I see NGDP growth as the benchmark, I still don’t consider money to have been very easy. The high oil prices didn’t spill over into core inflation. So we can agree to disagree on terminology, but on policy I will fiercely argue that money wasn’t too loose. That’s not much NGDP growth, and we now see how easy it was to fall into a liquidity trap.
calieconomist. If you look at the data there were clearly two housing crashes, each very different. The first occurred in 2007 and early 2008, and was concentrated in subprime markets. It missed many markets in the center of the US. By the spring of 2008 the fall in housing was starting to level off. Then around August a second and more severe crisis developed. This was not associated solely with subprime laons, but was caused by the general fall in NGDP in the US and world economies. Now even the central markets were all hit, as well as Canada, which avoided the first crash. And higher quality mortgages were hit. In addition, banks were hit by a huge drop in asset prices outside the residential real estate area. This hurt banks that had survived the initial subprime problems. And this second crash was caused by tight money.
2. August 2009 at 19:45
Perhaps you can point me to the data you refer to for the two separate crashes?
And even if there are two crashes, which caused the first? Tight money again, or something else?
Housing prices peaked in 2006-2007, and with home price appreciation, the bubble dependent economy went into recession.
Did tight money cause all housing bubbles to pop, or can they implode on their own? Money is powerful I agree, but it’s not everything.
3. August 2009 at 11:35
Calieconomist, The first was caused by many of the reasons everyone cites, it was not monetary policy, and I have no special expertise in that area. This link suggests the first crisis was easing before August 2008, which is when I think monetary policy caused the crisis to spread far beyond subprime markets to:
1. other housing markets
2. Industrial production
3. other countries, even ones with sound banking systems.
http://blogsandwikis.bentley.edu/themoneyillusion/?p=821#more-821
3. August 2009 at 16:37
Thank you for the link.
I find the increase in real interest rates in late 2008 interesting.
Citing two 5-month periods is unconvincing. The price deeclines accelerated. If you can look at the Case Shiller graphs from the past 5 years or so, it is clear housing largely peaked in 2006 and started declining. If you can find an uptick before August 2008, be my guest. And excluding the largest declines is transparent. You can’t exclude data that is inconvenient, sir.
http://www.newgeography.com/files/imagecache/Chart_fullnodeview/chartimages/Case-Shiller-Oct-08.png
Industrial production started falling in 2007 and the pace of decline increased in late 2008. Do you disagree with NBER that the recession started in 2007? If not, then the best your suggestions could obtain is a less severe recession.
http://research.stlouisfed.org/fred2/series/INDPRO
I agree that money demand increased in late 2008 increased, due to an increase in fear and risk aversion. Naturally it would be better if Bernanke had realized this, I agree. But it is misleading to call this “tight money”, which implies,to me, low money supply. Do you define tight money as high money demand? This seems disingenuous at best.
4. August 2009 at 06:27
Calieconomist,
I never said there was an “uptick” although if you use broader house price indexes that cover the entire country, not the Case Shiller index, there actually was a small uptick between February and August 2008. My point was that there were two very clear and separate declines in housing prices. Anyone who looks at a graph of the monthly rates of change (not levels) sees that immediately.
YOu said:
“And excluding the largest declines is transparent. You can’t exclude data that is inconvenient, sir.”
You obviously didn’t understand what I said, as I was merely discussing the spreed up in the decline after August. Of course there were larger declines earlier, if there weren’t my comment would have made no sense.
I would say that if the unemployment rate were still around 5.5% as in mid-2008, rather than 9.5%, as it is now, the recession would have been more than a bit milder. What would you say? In addition, if unemployment had never reached 6% the NBER would never have declared a recession (which they didn’t do until the end of 2008.) They don’t declare a recession until they are sure that it was bad enough to be called a recession. That’s why as of mid-2008 many economists still predicted that we would avoid a recession. The initial downturn was very mild. Indeed in its September 16, 2008 meeting the Fed (widely regarded as having an excellent forecasting unit) indicated that they thought that higher inflation was just as likely as recession.
The drops in industrial production before August 2008 were so mild as to be hardly worth commenting on. After August, IP plummeted. Many people wrongly believe that recessions are not avoidable once that have begun, that is completely false.
You said,
“I agree that money demand increased in late 2008 increased, due to an increase in fear and risk aversion. Naturally it would be better if Bernanke had realized this, I agree. But it is misleading to call this “tight money”, which implies,to me, low money supply. Do you define tight money as high money demand? This seems disingenuous at best.”
Use of monetary aggregates (or the base) as an indicator of monetary policy was almost totally discredited by the end of the 1980s. Obviously you are free to pick whatever indicator you wish, but most economists would not support your choice, nor would I. I do have a lot of posts that discuss the question of monetary policy indicators, if you are interested.
I define tight money as a monetary policy stance that is expected to produce less AD than the Fed wants, and vice versa. One can use market expectations (as I prefer) or the Fed’s expectations (as Lars Svensson prefers). By my definition money gradually got tight in the late summer. Using the Fed’s own forecast, tight money kicked in at the beginning of October.
4. August 2009 at 08:28
Calieconomist: I’m not an economist, so my non-expert summary may or may not be useful. With that: if the demand for something (including money) goes up substantially, and the supply doesn’t go up to match the demand, it seems reasonable to say that market is “tight”. The absolute levels don’t really matter.
4. August 2009 at 20:51
“My point was that there were two very clear and separate declines in housing prices. Anyone who looks at a graph of the monthly rates of change (not levels) sees that immediately.”
Well, all I see is declines. Using seasonally adjusted data, the best scenario is slightly less bad, i.e. slightly slower declines.
“You obviously didn’t understand what I said, as I was merely discussing the spreed up in the decline after August. Of course there were larger declines earlier, if there weren’t my comment would have made no sense.”
I apologize, I did misunderstand you. So housing price declines went from bad to worse, but why can’t that just be a worsening of the housing crisis?
“I would say that if the unemployment rate were still around 5.5% as in mid-2008, rather than 9.5%, as it is now, the recession would have been more than a bit milder. What would you say? In addition, if unemployment had never reached 6% the NBER would never have declared a recession (which they didn’t do until the end of 2008.) They don’t declare a recession until they are sure that it was bad enough to be called a recession.”
Yes it would be. But the 2001 recession never saw unemployment over 6%, so remaining under 6% doesn’t preclude a recession. They would have declared a recession, as things were getting worse. I would much prefer that recession to this one, that we agree on.http://research.stlouisfed.org/fred2/graph/?chart_type=line&s%5B1%5D%5Bid%5D=UNRATE&s%5B1%5D%5Brange%5D=10yrs
“Indeed in its September 16, 2008 meeting the Fed (widely regarded as having an excellent forecasting unit) indicated that they thought that higher inflation was just as likely as recession.”
Bernanke also said in 2007 that subprime was largely contained and that it wouldn’t affect the rest of the economy. Whoops.
http://www.forbes.com/2007/05/17/bernanke-subprime-speech-markets-equity-cx_er_0516markets02.html
I agree that your method is interesting and represents an improvement over current fed policy.
“The drops in industrial production before August 2008 were so mild as to be hardly worth commenting on. ”
Well, tell that to an unemployed industrial worker. I agree that things got much worse in August 2008, but a decline in industrial production signals recession.
“Use of monetary aggregates (or the base) as an indicator of monetary policy was almost totally discredited by the end of the 1980s.”
I am aware of this, which is why money cannot explain as much as monetarists believe.
Let’s take an example. NGDP is growing by 5% annually, let’s say 2% real growth and 3% inflation. Now an oil shock comes an reduces real GDP growth by 10% and increases inflation by 10%. NGDP is growing at 5%. What is the correct response in this case? Certainly not doing nothing.
I will check out the posts on monetary aggregates.
5. August 2009 at 10:37
Lawton, Well put.
Calieconomist.
I don’t have time to dig through all the old real estate data but it very clearly show two separate shocks–there is no ambiguity if you look at rates of change on a graph. The Economist magazine had a graph about 6 weeks back that showed this. It’s not just my imagination, and it’s not just seasonal factors. And it is much more obvious if you look beyond the C-S data, which is concentrated in subprime markets.
You said:
Yes it would be. But the 2001 recession never saw unemployment over 6%, so remaining under 6% doesn’t preclude a recession.
This is very misleading. Unemployment is a lagging indicator, and the rise over 6% in 2002 was clearly a part of the 2001 recession. I meant if it had stayed around 5.5% in the entire low cycle, not just the actual contraction. BTW, in meaningful economic terms the recession should be called over when the rate of growth returns to trend. I don’t like their dating approach. One percent RGDP growth is not “recovery” in any meaningful way, it is falling further behind the natural rate, or capacity, or whatever you call the trend line.
You said:
“Well, tell that to an unemployed industrial worker. I agree that things got much worse in August 2008, but a decline in industrial production signals recession.”
This isn’t true, small IP declines sometimes occur during boom periods. If they didn’t, then economists would have know we were in recession during the early months of 2008, but they didn’t. That’s because economists know that falling IP is often a false alarm.
You said:
“I am aware of this, which is why money cannot explain as much as monetarists believe.”
If by “money” you mean money supply (which is what the monetarists focused on), then I agree. If you mean monetary policy, then I disagree.
You said:
“Let’s take an example. NGDP is growing by 5% annually, let’s say 2% real growth and 3% inflation. Now an oil shock comes an reduces real GDP growth by 10% and increases inflation by 10%. NGDP is growing at 5%. What is the correct response in this case? Certainly not doing nothing.”
First of all that would never happen. But if it did, the correct approach would be to maintain 5% NGDP growth, as that best keeps the labor market in equilibrium. You want to avoid pushing the current nominal wage out of equilibrium, as with sticky wages that would worsen the business cycle.
I can’t even tell which way you think this would be wrong. Liberals tend to say “how can you allow RGDP to fall 10%.” Conservatives tend to say “How can you allow high inflation.” They are both wrong, NGDP targeting strikes the perfect balance. Which was your preferred alternative?
5. August 2009 at 13:42
Not to beat a dead horse, but perhaps you prefer OFHEO data for housing prices, as it is national in scope. I still see little slowing, but I guess we’ll have to disagree. http://www.fhfa.gov/Default.aspx?Page=14
“But if it did, the correct approach would be to maintain 5% NGDP growth, as that best keeps the labor market in equilibrium.”
This is a difficult and unlikely case, but my only point was that a rule needs to have some discretion for disastrous situations. The 5% NGDP target is interesting, but it would be somewhat inflexible. But other rule are similar, like naive inflation targeting, Taylor Rule, etc. My preferred alternative was the liberal one as a 10% drop in RGDP is unacceptable. I don’t think your choice was necessarily poor, I was just curious for your opinion.
6. August 2009 at 06:46
calieconomist, The problem with your example of a 10% drop in RGDP is that you assumed a supply shock. Monetary policy cannot prevent a supply shock from reducing RGDP, all one can do it to prevent the supply shock from raising unemployment. Now you might argue “how could a 10% fall in RGDP not raise unemployment?” That is why I said your example was far-fetched. If you were targeting NGDP it is hard to imagine any supply shock reducing RGDP 10%–it would have to be mind-blowingly bad. And if it were that bad, you couldn’t paper of the cost by printing money.
I was talking about graphs that show month over month rates of change. The Economist magazine had a good one about six or eight weeks ago in their “The World This week” business section. Maybe you can find in online. I think you will be surprised.